The iNaturalist Challenge 2017 Dataset

By Grant Van Horn, Oisin Mac Aodha, Yang Song, Alexander Shepard, Hartwig Adam, Pietro Perona, Serge J. Belongie. 2017

84 citations recorded by the source corpus.

Existing image classification datasets used in computer vision tend to have an even number of images for each object category. In contrast, the natural world is heavily imbalanced, as some species are more abundant and easier to photograph than others. To encourage further progress in challenging real world conditions we present the iNaturalist Challenge 2017 dataset - an image classification benchmark consisting of 675,000 images with over 5,000 different species of plants and animals. It features many visually similar species, captured in a wide variety of situations, from all over the world. Images were collected with different camera types, have varying image quality, have been verified by multiple citizen scientists, and feature a large class imbalance. We discuss the collection of the dataset and present baseline results for state-of-the-art computer vision classification models. Results show that current non-ensemble based methods achieve only 64% top one classification accuracy, illustrating the difficulty of the dataset. Finally, we report results from a competition that was held with the data.

Subfields: Computer Vision and Pattern Recognition, Ecological Modeling, Molecular Biology, Plant Science, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Ecological Modeling

In Molecular Biology

In Plant Science

In Evaluation & Benchmarks

Citations in the corpus

Toward Understanding Catastrophic Forgetting in Continual Learning (2019, background): “Achille et al. [2] empirically showed this effect on the iNaturalist dataset [53], where the distances between Task2Vec embeddings strongly agree with the distances be-tween natural taxonomical orders, hinting that the dissimilarity between tasks can be approximated from the distance between them…”
Grafit: Learning fine-grained image representations with coarse labels (2020, background): “It is preferable to pre-train on a domain similar to the target [16], e.g., pre-training on iNaturalist [29] is preferable to pre-training on ImageNet if the ﬁnal objective is to discriminate between species of birds.”
Recognition in Terra Incognita (2018, background): “Natural world datasets such as the iNaturalist dataset [1], CUB200 [32], Oxford Flowers [33], LeafSnap [34], and NABirds700 [35] are focused on finegrained species classification and detection.”
Multi-Attention Multi-Class Constraint for Fine-grained Image Recognition (2018, background): “Recently, there are several attempts such as Goldfinch [19] and the iNaturalist Challenge [38] in building large-scale fine-grained benchmarks.”
Learning with Imprinted Weights (2017, methodology): “As a test of imprinting on a large-scale and diverse dataset, we apply imprinting to the learning of novel categories on the iNaturalist dataset [21].”
A computer vision for animal ecology. (2018, background): “The natural excitement for plants and animals means that gathering further labelled data is possible through online citizen scientist efforts (Van Horn et al., 2015).”
Feature Transfer Learning for Face Recognition With Under-Represented Data (2019, background): “While this paper focuses on face recognition, our future work will also derive advantages from the proposed feature transfer for other recognition applications, such as UR natural species [17].”
Learning to Zoom: a Saliency-Based Sampling Layer for Neural Networks (2018, methodology): “In this experiment, we study the problem using the iNaturalist dataset that contains 5,089 animal species [29].”
CropDeep: The Crop Vision Dataset for Deep-Learning-Based Classification and Detection in Precision Agriculture (2019, background): “VegFru is a new domain-speciﬁc dataset for ﬁne-grained visual classiﬁcation (FGVC) [23].”
Machine learning for image based species identification (2018, background): “Examples highlighting the wealth of images captured by researchers and the public are the iNaturalist (675,000 images of 5,000 species, Van Horn et al., 2017) and the Zooniverse (1.2 million images of 40 species; Swanson et al., 2015) databases.”
Coarse2Fine: a two-stage training method for fine-grained visual classification (2019, background): “One less studied but challenging dataset in FGVC is iNaturalist [6].”
GraphSMOTE: Imbalanced Node Classification on Graphs with Graph Neural Networks (2021, background): “Plenty of tasks suffer from this problem, like medical diagnosis [13, 24] or fine-grained image classification [29, 36].”
The iNaturalist Species Classification and Detection Dataset (2017, background): “they only differ in terms of pose, lighting, color, but not necessarily in their underlying object shape or appearance [49, 7, 50].”